What Fluidstack does

Fluidstack is a UK-incorporated, New York-headquartered neocloud that builds and operates gigawatt-scale AI compute campuses. The company acquires power, designs and constructs data centers, deploys HPC clusters, and operates them under a vertically integrated stack that the company claims brings campuses online in approximately six months versus the 18-24 month industry norm. Its product surface spans the Fluidstack AI Cloud Platform, the Fluidstack Console software control plane for provisioning and managing HPC capacity, and end-to-end Data Center Infrastructure Services covering power, design, build, and operations.

Fluidstack serves AI labs and AI-adjacent enterprises that require very large, sustained compute capacity. Its flagship relationship is a multi-year, multi-billion-dollar compute build-out agreement with Anthropic spanning Texas, New York, and Louisiana campuses, supported by Google as a financial backstop. The customer base also includes Meta, OpenAI, Poolside, Black Forest Labs, and Boost Run. Distribution is hybrid: long-dated enterprise field-sales contracts with frontier AI labs are paired with self-serve provisioning via the Fluidstack Console, while physical capacity is delivered through Fluidstack-operated campuses in partnership with site operators (Hut 8, TeraWulf, Cipher Digital) and via JV vehicles with Coatue/Next Frontier.

The revenue model combines multi-year subscription Orders (auto-renewing with advance payment) with per-minute/per-hour usage-based metering on top, plus bespoke multi-year, multi-billion-dollar build-out contracts. Pricing is quote-based and not publicly disclosed. The company has scaled rapidly: it secured a $200M Series A in February 2025, a ~$750M round at a $7.5B valuation by January 2026, and entered talks in April 2026 for an additional ~$1B at an $18B valuation; concurrent activity includes the Broadcom/Apollo/Blackstone AI XPV Platform deployment (Anthropic as first customer), a Google-backed $5.7B senior secured note financing at the Indiana JV, and a withdrawal from European projects to focus exclusively on the U.S. market.

Name

Fluidstack

Legal name

Fluidstack Limited (company number 10985545)

Website

https://fluidstack.io

Company type

Private

Founded year

2017

Operating status

Operating

Headcount range

101–250 employees

Short description

Fluidstack is a UK-incorporated, US-headquartered neocloud that builds and operates gigawatt-scale AI compute campuses for frontier AI labs, notably under a multi-billion-dollar Anthropic build-out, using a vertically integrated power-to-console stack that delivers campuses in roughly six months.

Ownership category

akta.pro rank

Product category

AI Cloud Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210), Computer Facilities Management Services (541513), Computer Systems Design and Related Services (5415)

SIC

Services-Computer Integrated Systems Design (7373), Services-Computer Processing & Data Preparation (7374), Services-Computer Programming, Data Processing, Etc. (7370), Water, Sewer, Pipeline, Comm & Power Line Construction (1623)

akta.pro primary industry

AI Compute Cloud & GPU-as-a-Service (HDAAAAAK)

akta.pro secondary industries

GPU-Accelerated & AI Training/Inference Servers (HDACABAG), AI/HPC Converged Systems (Hybrid Training/Simulation) (HDACAOAD), HPC Cooling, Power & Infrastructure (Liquid Cooling, Racks, Power Delivery) (HDACAOAI), AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG), End-to-End MLOps & ML Platform Suites (HDAAABAA), AI/ML Solution Integration & MLOps Enablement (BPAEAAAJ)

Keywords

• GPU cloud infrastructure
• AI compute platform
• Neocloud services
• Data center colocation
• HPC cluster deployment

Headquarters

HQ city

New York

HQ country

United States

HQ region

North America

Offices10 records

Markets served

GTM type

B2B

Offering type

Services

Cost components

Infrastructure, Technology or R&D, Personnel, Operations, Marketing or Sales

Revenue model

1. Compute subscriptions and Orders: Recurring revenue under Customer Orders, which auto-renew unless terminated. Customers are billed for HPC compute capacity on a subscription-style basis per the Terms & Conditions.
2. Usage-based HPC compute consumption: Customers are charged on a per-minute / per-hour basis for active compute usage in addition to subscription fees, with auto-renewing Orders and advance-payment terms.
3. Multi-year, multi-billion-dollar build-out contracts: Custom-built, long-duration infrastructure deals with frontier AI labs (e.g., multi-billion-dollar agreement with Anthropic) representing large, contractually committed revenue tied to specific gigawatt-scale deployments.
4. Managed data center / colocation services: Fluidstack operates sites such as Lake Mariner and River Bend, providing shell, power and HPC operations to AI customers (including TPU and GPU tenants) under long-term arrangements.

Pricing tiers

Go-to-market motion4 records

Distribution channels3 records

Marketing channels4 records

Core offering

Fluidstack builds and operates gigawatt-scale AI compute infrastructure for training and inference workloads. The company acquires power, designs and constructs data centers, and runs HPC clusters with both NVIDIA GPUs and Google TPUs, exposing that capacity through long-dated contracts to frontier AI labs (Anthropic, Meta, OpenAI, Poolside, Black Forest Labs) and through a self-serve software control plane, the Fluidstack Console, on subscription plus usage-based terms. End-to-end service spans power procurement, site/construction, and ongoing operations at campuses such as River Bend (245MW), Lake Mariner (430MW+), Barber Lake (300MW) and a 430MW Indiana JV.

Product overview

Fluidstack offers a unified AI cloud infrastructure platform that combines three integrated pillars: (1) the Fluidstack AI Cloud Platform, which delivers gigawatt-scale GPU (NVIDIA) and TPU (Google) compute capacity for AI training and inference workloads for customers like Anthropic, Meta, Poolside, Black Forest Labs, and Boost Run; (2) the Fluidstack Console, a self-service software platform accessible at https://console2.fluidstack.io/ that serves as the primary management interface for placing orders, managing services, and handling subscriptions; and (3) end-to-end Data Center Infrastructure Services spanning power acquisition, site selection, design, construction, commissioning, and operations. The licensed Fluidstack Software ties these pillars together, enabling customers to access the underlying compute resources. Together, these offerings form a vertically integrated stack where Fluidstack handles everything from power procurement to the user-facing console, positioning itself as a neocloud alternative to traditional hyperscalers with a focus on faster deployment timelines (6 months vs. 18-24 months industry standard).

Differentiator

Problem solved

Functional benefit

Products and services

• Fluidstack AI Cloud Platform Gigawatt-scale GPU and TPU compute platform for AI training and inference, contracted via multi-year build-out agreements and subscription Orders. Customers include Anthropic, Meta, Poolside, Black Forest Labs, Boost Run, Core42, and OpenAI.
• Fluidstack Console Self-serve software control plane for placing Orders, managing Services, monitoring usage, and handling subscriptions for Fluidstack HPC compute. Customers enter Orders that auto-renew and are billed per-minute/per-hour plus subscription fees, with advance-payment terms.
• Data Center Infrastructure Services End-to-end build-and-operate services covering power acquisition, site selection, design, construction, commissioning and ongoing operations for hyperscale AI workloads. Fluidstack claims to deliver gigawatts of compute in roughly 6 months versus the 18-24 month industry standard, with active projects at Lake Mariner (NY), River Bend (LA), Barber Lake (TX), Abernathy (TX), and New Lebanon (IN).

Quantifiable outcome

• Delivers gigawatt-scale compute campuses in ~6 months vs. industry 18-24 month norm.
• +4 more outcomes

Named customers7 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration2 records

AI capability2 records

Feature3 records

Partnerships

Nine partnerships are on record, tiered flagship, core and major.

• AnthropicflagshipStrategic or Co-development Partner · 1 January 2025Multi-year, multi-billion-dollar compute build-out agreement under which Fluidstack is constructing large-scale infrastructure dedicated to Anthropic. Reported as one of the largest AI-infrastructure commitments to date and a centerpiece of Fluidstack's revenue backlog.
• Hut 8coreStrategic or Co-development Partner · 1 January 2025Long-term, ~$7B lease arrangement at Hut 8's River Bend campus in Louisiana for a 245MW AI compute site, with Fluidstack as primary tenant and HPC operator.
• TeraWulfcoreStrategic or Co-development PartnerPartnership covering the Lake Mariner site in New York, where TeraWulf provides infrastructure and Fluidstack delivers HPC compute capacity, including hosting Core42 and other AI customers.
• Cipher DigitalcoreStrategic or Co-development PartnerLease and build-out arrangement at Cipher Digital's Barber Lake site in Texas for Fluidstack-operated AI compute capacity.
• Core42majorStrategic or Co-development PartnerCo-tenant / capacity partner at the Lake Mariner site, taking HPC capacity from Fluidstack-operated infrastructure.
• Coatue Management / Next FrontiermajorStrategic or Co-development PartnerJoint venture with Next Frontier (backed by Coatue) via Meridian Arc HoldCo to develop an Indiana data center campus for Fluidstack-operated AI capacity.
• Meridian Arc HoldComajorStrategic or Co-development PartnerJV vehicle formed with Next Frontier / Coatue for the Indiana data center project, holding the campus real estate and power rights for Fluidstack-operated compute.
• BroadcommajorTechnology or IntegrationChip supply and integration partner supporting Fluidstack's HPC build-outs.
• Schneider ElectricmajorTechnology or IntegrationInfrastructure vendor supporting power, cooling and electrical systems across Fluidstack data center campuses.

Scale indicators9 records

Recent moves7 records

Expansion highlights6 records

Direct peers

• CoreWeave: CoreWeave is the closest direct peer: a neocloud that delivers large-scale NVIDIA GPU compute to frontier AI labs and enterprises, including similar anchor customers. It competes head-to-head with Fluidstack for multi-gigawatt AI training deals and IPO'd at a multibillion-dollar valuation on the same thesis.
• Lambda Labs: Lambda operates a GPU cloud focused on AI training and inference, serving AI labs and enterprises with NVIDIA hardware. Like Fluidstack, it offers both on-demand and reserved GPU compute and targets the same neocloud buyer, though historically at smaller deployment scale.
• Crusoe Energy: Crusoe pivoted from flared-gas crypto mining into AI compute and now operates large modular data centers powered by stranded energy, mirroring Fluidstack's strategy of using power-advantaged sites. It sells GPU compute to AI startups and labs and competes directly for similar frontier AI workloads.
• Nebius Group: Nebius is a publicly listed AI infrastructure company that builds and operates GPU clusters and a full-stack AI cloud platform for model builders. It competes with Fluidstack for large-scale AI training customers and shares the vertically integrated 'build-and-operate' positioning.

Emerging players

• Applied Digital: Applied Digital pivoted from crypto mining to AI/HPC colocation and now designs, builds and operates large AI-ready data centers. It is a smaller but rapidly scaling competitor to Fluidstack in the AI build-to-suit data center space.
• Together AI: Together AI is an AI cloud platform built around open-source and frontier model inference and training, operating its own GPU clusters. It overlaps with Fluidstack's Fluidstack Console-style self-serve AI cloud motion at smaller scale.

Broad incumbents

• Vultr: Vultr is a global cloud infrastructure provider offering GPU instances alongside general-purpose compute and hosting. It competes with Fluidstack on the self-serve / developer side of the HPC cloud market, but at much smaller per-customer scale and without gigawatt builds.
• Equinix: Equinix is a global colocation and interconnection incumbent whose data centers and cross-connect fabric host a large share of enterprise and AI workloads. It competes with Fluidstack for enterprise AI deployments, but as a broader-portfolio colocation operator rather than a specialist AI neocloud.
• Digital Realty Trust: Digital Realty is a global data center REIT that builds and operates hyperscale and colocation facilities, including a growing footprint of AI-ready builds. It is comparable to Fluidstack in providing physical AI infrastructure at scale, but operates a REIT model rather than direct AI cloud services.
• OVHcloud: OVHcloud is a European-headquartered cloud and bare-metal / dedicated server provider that competes in the AI/HPC hosting market. It is comparable to Fluidstack on the broader 'sovereign / alternative-to-hyperscaler' cloud axis but at much smaller per-deployment scale.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/fluidstack
• LinkedInlinkedin.com/company/fluidstack

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Subsidiaries3 records

Funding overview

Funding rounds13 records

Investors12 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Fluidstack